At a conference of more than 3,500 users, Splunk executives showed off their company’s latest tools. Splunk makes software for discovering, monitoring and analyzing machine data, which is often considered data exhaust since it is a by-product of computing processes and applications. But machine data is essential to a smoothly running technology infrastructure that supports business process. One advantage is that because machine data not recorded by end users, it is less subject to input error. Splunk has grown rapidly by solving fundamental problems associated with the complexities of information technology and challenging assumptions in IT systems and network management that is rapidly being referred to as big data analytics. The two main and related assumptions it challenges are that different types of IT systems should be managed separately and that data should be modeled prior to recording it. Clint Sharp, Splunk’s director of product marketing, pointed out that network and system data can come from several sources and argued that utilizing point solution tools and a “model first” approach does not work when it has to deal with big data and a question-and-answer paradigm. Our research into Operational Intelligence finds that IT systems are most important information source in almost two thirds (62%) of organizations. Splunk used the conference to show how it has brought to these data management innovations the business trends of mobility, cloud deployment and security.
Presenters from major customer companies demonstrated how they work with Splunk Enterprise. For example, according to Michael Connor, senior platform architect for Coca-Cola, bringing all the company’s data into Splunk allowed the IT department to reduce trouble tickets by 80 percent and operating costs by 40 percent. Beyond asserting the core value of streamlining IT operations and the ability to quickly provision system resources, Connor discussed other uses for data derived from the Splunk product. Coca-Cola IT used a free community add-on to deliver easy-to-use dashboards for the security team. He showed how channel managers compare different vending environments in ways they had never done before. They also can conduct online ethnographic studies to better understand behavior patterns and serve different groups. For Coca-Cola, the key to success for the application was to bring data from various platforms in the organization into one data platform. This challenge, he said, is more to do with people and processes than technology, since many parts of an organization are protective of their data, in effect forming what he called “data cartels.” This situation is not uncommon. Our research into information optimization shows that organizations need these so-called softer disciplines to catch up with their capabilities in technology and information to realize full value from data and analytics initiatives.
In keeping up with trends, Splunk is making advances in mobility. One is MINT for monitoring mobile devices. With the company’s acquisition of BugSense as a foundation, Splunk has built an extension of its core platform that consumes and indexes application and other machine data from mobile devices. The company is offering the MINT Express version to developers so they can build the operational service into their applications. Similar to the core product, MINT has the ability to track transactions, network latency and crashes throughout the IT stack. It can help application developers quickly solve user experience issues by understanding root causes and determining responsibility. For instance, MINT Express can answer questions such as these: Is it an application issue or a carrier issue? Is it a bad feature or a system problem? After it is applied, end-user customers get a better digital experience which results in more time spent with the application and increased customer loyalty in a mobile environment where the cost of switching is low. Splunk also offers MINT Enterprise, which allows users to link and cross-reference data in Splunk Enterprise. The ability to instrument data in a mobile environment, draw a relationship with the enterprise data and display key operational variables is critical to serving and satisfying consumers. By extending this capability into the mobile sphere, Splunk MINT delivers value for corporate IT operations as well as the new breed of cloud software providers. However, Splunk risks stepping on its partners’ toes as it takes advantage of certain opportunities as in mobility. In my estimation, the risk is worth taking given that mobility is a systemic change that represents enormous opportunity. Our research into business technology innovation shows mobility in a virtual tie with collaboration for the second-most important innovation priority for companies today.
Cloud computing is another major shift that the company is prioritizing. Praveen Rangnath, director of cloud product marketing, said that Splunk Cloud enables the company to deliver 100 percent on service level agreements through fail-over capabilities across AWS availability zones, redundant operations across indexers and search heads, and by using Splunk on Splunk itself. Perhaps the most important capability of the cloud product is its integration of enterprise and on-demand systems. This capability allows a single view and queries across multiple data sources no matter where they physically reside. Coupled with Splunk’s abilities to ingest data from various NoSQL systems – such as Mongo, Cassandra, Accumulo, Amazon’s Elastic Map Reduce, Amazon S3 and even mainframes – with the Ironstream crawler, its hybrid search capability is unique. The company’s significant push into the cloud is reflected by both a 33 percent reduction in price and its continued investment into the platform. According to our research into information optimization one of the biggest challenges with big data is simplification of data access; as data sources increase easy access becomes more important. More than 92 percent of organizations that have 16 to 20 data sources rated information simplification very important. As data proliferates both on-premises and in the cloud, Splunk’s software abstracts users from the technical complexities of integrating and accessing the hybrid environment. (Exploring this and related issues, our upcoming benchmark research into data and analytics in the cloud will examine trends in business intelligence and analytics related to cloud computing.)
Usability is another key consideration: In our research on next-generation business intelligence nearly two-thirds (63%) of organizations said that is an important evaluation criterion, more than any other one. At the user conference Divanny Lamas, senior manager of product management, discussed new features aimed at the less sophisticated Splunk user. Advanced Feature Extractor enables users to extract fields in a streamlined fashion that does not require them to write an expression. Instant Pivot enables easy access to a library of panels and dashboards that allows end users to pivot and visually explore data. Event Pattern Detection clusters patterns in the data to make different product usage metrics and issues impacting downtime easier to resolve. Each of these advances represents progress in broader usability and organizational appeal. While Splunk continues to make its data accessible to business users, gaining broader adoption is still an uphill battle because much of the Splunk data is technical in nature. The current capabilities address the technologically sophisticated knowledge worker or the data analyst, while a library of plug-ins allows more line-of-business end-users to perform visualization. (For more on the analytic user personas that matter in the organization and what they need to be successful, please see my analysis.)
Splunk is building an impressive platform for collecting and analyzing data across the organization. The question from the business analytics perspective is whether the data can be modeled in ways that easily represent each organization’s unique business challenges. Splunk provides search capabilities for IT data by default, but when other data sources need to be brought in for more advanced reporting and correlation, it requires the data to be normalized, categorized and parsed. Currently, business users apply various data models and frameworks from major IT vendors as well as various agencies and data brokers. This dispersion could provide an opportunity for Splunk to provide a unified platform; the more data businesses ingest, the more likely they will rely on such a platform. Splunk’s Common Information Model provides a metadata framework using key-value pair representation similar to what other providers of cloud analytic applications are doing. When we consider the programmable nature of the platform including RESTful APIs and various SDKs, HUNK’s streamlined access to Hadoop and other NoSQL sources, Splunk BD connect for relational sources, the Splunk Cloud hybrid access model and the instrumentation of mobile data in MINT, the expansive platform idea seems plausible.
A complicating factor as to whether Splunk will become such a platform for operational intelligence and big data analytics is the Internet of Things (IoT), which collects data from various devices. Massive amounts of sensor data already are moving through the Internet, but IoT approaches and service architectures are works in progress. Currently, many of these architectures do not communicate with others. Given Splunk’s focus on machine data which is a key type of input for big data analytics in 42 percent of organizations according to our research, IoT appears to be a natural fit. IoT is generating event-centered data which is a type of input for big data analytics in 48 percent of organizations. There is some debate about whether Splunk is a true event processing engine, but that depends on how the category is defined. Log messages, its specialty, are not events per se but rather are data related to something that has happened in an IT infrastructure. Once correlated, this data points directly to something of significance, including events that can be acted upon. If such a correlation triggers a system action, and that action is taken in time to solve the problem, then the data provides value and it should not matter if the system is acting in real time or near real time. In this way, the data itself is Splunk’s advantage. To be successful in becoming a broader data platform, the company will need to advance its Common Information Model, continue to emphasize the unique value of machine data, build their developer and partner ecosystem, and encourage customers to push the envelope and develop new use cases.
For organizations considering Splunk for the first time, IT operations, developer operations, security, fraud management and compliance management are obvious areas to evaluate. Splunk’s core value is that it simplifies administration, reduces IT costs and can reduce risk through pattern recognition and anomaly detection. Each of these areas can deliver value immediately. For those with a current Splunk implementation, we suggest examining use cases related to business analytics. Specifically, comparative analysis and analysis of root causes, online ethnography and feature optimization in the context of the user experience can all deliver value. As ever more data comes into their systems, companies also may find it reasonable to consider Splunk in other new ways like big data analytics and operational intelligence.